Publication number | US7006680 B2 |
Publication type | Grant |
Application number | US 10/430,664 |
Publication date | Feb 28, 2006 |
Filing date | May 5, 2003 |
Priority date | May 3, 2002 |
Fee status | Paid |
Also published as | DE60323625D1, EP1502098A1, EP1502098B1, US7466851, US20030215867, US20050105787, WO2003093810A1 |
Publication number | 10430664, 430664, US 7006680 B2, US 7006680B2, US-B2-7006680, US7006680 B2, US7006680B2 |
Inventors | Sandeep Gulati |
Original Assignee | Vialogy Corp. |
Export Citation | BiBTeX, EndNote, RefMan |
Patent Citations (56), Non-Patent Citations (3), Referenced by (31), Classifications (22), Legal Events (5) | |
External Links: USPTO, USPTO Assignment, Espacenet | |
This invention claims priority from U.S. patent application Ser. No. 60/377,520, filed on May 3, 2002, which is hereby fully incorporated by reference.
A Software Appendix comprising four sheets is included herewith.
The invention relates to a system and method for characterizing output data from an arrayed platform, and more specifically, arrays for characterizing biological material, including genetic material.
Arrayed platforms or arrays typically consist of an arrangement of two or more detectors in one or more dimensions. The detectors of such arrays output data that may be monitored to observe various conditions of interest. In particular, arrayed platforms with a plethora of detectors may be useful in simultaneously monitoring a wide range of conditions by providing many different detectors and/or by utilizing numerous identical detectors for redundant measurements. However, errors still occur when characterizing data that is output from an array.
Most arrayed platforms suffer from background noise and other interferences that distort or obscure the output data resulting in erroneous measurements. This is particularly the case where the intensity of the output from each detector is close to or below the background noise threshold.
One type of arrayed platform are microarrays which typically comprise of a plurality of detectors (target probes) to which biological matter, such as a sample of genetic material (e.g., oligonucleotides, peptides, cDNA, proteins, etc.) may hybridize in some fashion to complimentary material. The genetic material used with typical microarray experiments is often labeled using fluorescence, chemiluminescence, bioluminescence, photoluminescence, or some other spectral emitter. The presence of a spectral emitter at a detector probe (also referred to herein as an object of interest) suggests that hybridization has taken place, and therefore, that the sample includes genetic material corresponding to the detector. However, the background noise on microarrays, label signals associated with non-specific hybridization, poor resolution, and protocol variations, among other things, often distort the spectral output and result in false negatives where an expressed hybridization signal is not detected, and false positives, where noise, platform defects, or other signals (such as non-specific hybridization) are incorrectly determined to be signals associated with hybridization.
The spectral intensity generated by a microarray is detected using a scanner (e.g., laser scanner, CCD array) or other detection device such as those sold by Axon, Affymetrix, Agilent and others. The detected spatial intensity is saved as a pixellated or raster output pattern data file (referred to herein as an “output pattern”). Typically, these images are saved in standard formats such as .jpg, .tif, gif, etc. and may be analyzed using a variety of analysis programs. The output patterns comprise an array of contiguous pixels that include both the pixellated intensity provided by the detectors on the array as well as other portions of the array within the total scan area of the scanner.
The total intensity emitted by an object of interest (i.e., a detector) within an output pattern is commonly referred to as a “feature” or “spot”. One or more features or spots may correspond to a specific gene that is being detected or analyzed. This identification may occur by parameterizing the spots, using spot size, spacing, layout, packaging, distortion boundaries, etc.
Traditional microarray characterization techniques analyze pixels associated with objects of interest, as well as pixels associated with background noise on the microarray. However, such techniques are imprecise and require the analysis of large amounts of data that is not germane to the microarray experiment. In addition, while some techniques attempt to extract relevant information from an output pattern, conventional systems often exclude pixels associated with an object of interest and/or include pixels associated with background results, resulting in inaccurate and imprecise characterization (especially with regard to microarrays used for weakly expressed gene detection and gene quantitation). The imprecision in results becomes worse for situations where (i) images are very noisy due to instrument, coating, linker chemistry, or platform substrate noise; (ii) images are noisy due to biological or assay noise; (iii) the sample applied to the microarray is inadequately labeled due to poor labeling efficacy; (iv) photo bleaching effects; (v) imprecise and variations in spotting; (vi) labels or dyes bleed through in images where multiple dye scanning are used in connection with energy transfer dyes; and (vii) the output patterns often do not have sufficient spatial resolution to characterize each object of interest on an array.
Conventional techniques for spot extraction utilized in hybridization images (microarrays, biochips and protein arrays), include rectilinear scanning on regular grids and the use of extraction masks. Rectilinear scanning refers to a pixellated spot represented in Cartesian (x, y) coordinate space. With this arrangement, the spot, covering m×n pixels, is either traversed in x-direction, row-by-row for the m-rows, or it is traversed in the y-direction for all the n-columns. With extraction masks, a square, circular, elliptical or polygonal mask is pre-designed as an extraction template. A “set-intersection” with the pre-configured mask is then performed to extract all the pixels that are contained within the mask boundary. The mask application may be preceded by an additional spot registration step that entails sliding the mask around the centroid of the segmented object to a position that minimizes the number of segmented pixels that are outside the mask boundary. Alternatively, the segmented object may be translated in a rectilinear or diagonal direction to increase precision. While rectilinear scanning and extraction masks provided an advantage over some extraction techniques, there remains a need for an improved system with increased precision that is adaptable to extraction non-standard spot sizes. In addition, there remains a need for an array analysis system that can provide enhanced spatial resolution over output patterns generated by a scanning/detection device.
Notwithstanding the above, arrayed platforms remain a useful and promising technique for simultaneously detecting multiple conditions of interest. Greater adoption of arrays would occur if there was a system and method for characterizing output patterns that was robust and repeatable. Accordingly, it will be appreciated that there remains a need for an improved system for extracting information relating to an object of interest within an array output pattern and for enhancing the resolution of an array output pattern. It is to these and other ends that aspects of the present invention are primarily directed.
The invention is disclosed, in a technique for characterizing information from a pixellated output pattern of a microarray having one or more objects of interest. The technique commences with the step of extracting pixels within the output pattern representative of each object of interest. Next, the intensity representation of the extracted pixels are transformed to a spectral representation.
In some embodiments, the step of transforming the intensity representations of the extracted pixels is dependent upon the expected signal level for each object of interest and may be adjusted accordingly.
In some embodiments, the pixels are extracted using a radial spiral transversal. With some arrangement the radial spiral traversal originates at a centroid of each object of interest and progresses outwardly and in other arrangements the radial spiral traversal originates at a pre-determined distance from the centroid. The radial spiral traversal may terminate at a segment boundary pixel for each object of interest, or the radial spiral traversal may terminate at a pre-determined distance beyond a segment boundary for each object of interest. In yet another embodiment, the radial spiral traversal terminates at a pre-determined distance before a segment boundary for each object of interest.
In yet other embodiments, the radial spiral traversal converges inwardly toward a centroid of each object of interest. With such arrangements, the radial spiral traversal may terminate at the centroid for each object of interest, or it may terminate at a pre-determined distance inside a segment boundary for each object of interest. The radial spiral traversal may originate on a segment boundary pixel, at a pre-determined distance outside a segment boundary, or within a segment boundary for each object of interest.
The use of logarithmic spirals are also disclosed to extract pixels. In some embodiments, the logarithmic spiral uses a tessellated extraction mask, which may comprise a texture mask. A variety of texture masks may be used in connection with the current invention including Bernoulli spiral masks, logistique masks, Hirschorn masks, Voderberg masks, bent-wedge tile masks, kinked tile masks, rhomboidal tile masks, triangular tile masks, equiangular spiral masks, symmetric tessellation masks, asymmetric tessellation masks, and spiral mirabilis masks.
The current invention also may adapt to differing types of objects of interest. For example, the radial spiral traversal for pixel extraction may be configured to extract pixels associated with objects of interest having varying geometric sizes, pixels associated with objects of interest distributed geometrically irregularly on the microarray, and pixels associated with objects of interests having varying geometric sizes and distributed geometrically irregularly on the microarray.
Other types of pixel extraction techniques that may be used in connection with the current invention include degenerate logarithmic spiral masks, rectilinear masks, row-major extraction, column-major extraction.
In some variation of the preferred embodiment, the characterization technique further comprises the step of computationally re-scaling the extracted pixels to increase spatial resolution of the extracted pixels prior to the step of transformation the intensity representation. The computationally re-sampling the extracted pixels comprises may result from the convolving of the extracted pixels with a kernel, which may be a discretized kernel with coefficients chosen from the groups comprising linear functions, non-linear functions, and canonical kernel functions. In some embodiments, the computational re-sampling of the extracted pixels results from the product of a scalar dot product with an affine transformation, or a linear function. The spatial resolution may also be increased by convolution of the extracted pixels with a canonical coefficient kernel that yields two or more modified pixels for each extracted pixel, or through the utilization of a cascade of computational convolutions with discretized canonical kernel functions, with each successive convolution conducted on the results of preceding convolution. Such convolution cascades may be serial cascades, parallel cascades, or a a combination of a serial cascade and a parallel cascade.
The output pattern that is characterized may comprise an array of spatially contiguous pixels. In such arrangements, it is determined which pixels are representative of each object of interest, or indicative of morphological representation of an object of interest. Information-theory or information measure may be used to determine which pixels should be associated with an object of interest.
The invention discloses the use of the step transforming the domain of the extracted pixels from spatial intensity to spatial frequency. This transformation may occur through a Fourier transform, such as a Fast Fourier Transform, or a Discrete Fourier Transform. Once the domain has been transformed to a spatial representation, a spectral vector representing the transformed object may be partitioned into overlapping, or non-overlapping subvectors prior to convolving with discretized coefficients of a function. The power spectral density may be estimated from the spectral representation of the extracted pixels. In some embodiments, a spectrally transformed vector is decomposed into subvectors associated with an object of interest, where each subvector is convolved with a resolution enhancement kernel and synthetically resampled. The post-convolution computationally re-sampled transformed spectral subvectors may then be combined to yield a single spectral vector.
In some embodiments of the current invention, the technique further includes the step of decomposing the extracted pixels into a set of discrete extracted objects, where each discrete extracted object is transformed into a spectral representation. In other embodiments, the technique removes pixels that do not pertain to an object of interest prior to the pixel extraction step.
The invention also discloses the use of a logarithmic spiral to estimate a local background of an object of interest
A wide variety of arrays may be utilized with respect to the current invention, including arrays such as hybridized spotted cDNA microarrays, synthesized oligonucleotide arrays, spotted oligonucleotide arrays, peptide nucleotide assays, single nucleotide polymorphism (SNP) arrays, carbohydrate arrays, glycoprotein arrays, protein arrays, proteomic arrays, tissue arrays, antibody arrays, antigen arrays, bioassays, sequencing microarrays, sequencing by hybridization (SBH) microarrays, siRNA duplexes, RNAi arrays glass-based arrays, nylon membrane arrays, thin film arrays, polymer-substrate arrays, capillary electrophoresis arrays, genospectral arrays, electronic arrays, bead arrays, quantum dot arrays, and gylcan arrays. Such arrays may include emitters including: fluorescence, chemiluminescence, bioluminescence, and photoluminescence.
In order to facilitate the approximate location of objects of interest, the technique includes the step of registering and/or segmenting the pixels within the output pattern to approximate the location of the objects of interest.
In some analysis techniques, it may also be desirable to adjust the contrast of the output pattern. The invention describes a variety of contrast filters including Gabor filters, low-pass band-pass filters, high-pass band-pass filters, edge detection operators, Laplacian filters, gradient-focusing filters.
Another specific method is disclosed which may be utilized to characterize information from a pixellated microarray output pattern. This method includes the steps of segmenting the output pattern to approximate the locations of the objects of interest within the output pattern; filtering the segmented output pattern to enhance the contrast; extracting a core of pixels associated with each object of interest; resampling the extracted core pixels to enhance spatial resolution; transforming the representation of the resampled extracted core pixels from the intensity domain to the spectral domain; convolving the transformed spectral domain resampled extracted core pixels with a kernel; and estimating the power spectral density from the convolved transformed spectral domain resampled extracted core pixels.
The current invention is also embodied in a computer code product that characterizes information from a pixellated output pattern of microarray that represents one or more objects of interest, The computer code product includes computer code that extracts spatial data elements within the output pattern representative of each object of interest; and computer code that transforms the intensity representation of the extracted pixels to a spectral representation
The current invention is also disclosed in a computer system for characterizing information from a pixellated output pattern of a microarray representing one or more objects of interest. The computer system includes a processor and a memory coupled to said processor. The memory is configured to encode one or more programs that may perform the steps of: extracting spatial data elements within the output pattern representative of each object of interest; and transforming the intensity representation of the extracted pixels to a spectral representation.
In yet another embodiment, a system for analyzing an output pattern of an array of detectors, each detector representing one or more objects of interest is disclosed. This system includes means for extracting spatial data elements within the output pattern representative of each object of interest; and means for transforming the intensity representation of the extracted pixels to a spectral representation.
The current invention is also disclosed as an article of manufacture, made by a method for characterizing information from a pixellated output pattern of an array of detectors, each detector representing one or more objects of interest. The method includes the steps of extracting spatial data elements within the output pattern representative of each object of interest; and transforming the intensity representation of the extracted pixels to a spectral representation.
As can be appreciated, a wide range of embodiments are consistent with the general principles of the invention.
FIG. 6(a) illustrates an exemplary logarithmic spiral trajectory overlaid on a pixellated grid;
FIG. 6(b) illustrates the ordering of the pixels as traversed in FIG. 6(a);
FIG. 8(a) illustrates a sample logarithmic spiral with a pre-defined parametric constant a having a value of 0.15;
FIG. 8(b) illustrates a sample logarithmic spiral with a pre-defined parametric constant a having a value of 0.17;
FIG. 8(c) illustrates a sample logarithmic spiral with a pre-defined parametric constant α having a value of 0.25;
FIG. 9(a) illustrates a single object of interest overlaid on a grid;
FIG. 9(b) illustrates a first embodiment of a tessellated spiral extraction mask overlaying an object of interest with outward radial progression commencing from the centroid of the object of interest;
FIG. 9(c) illustrates a first embodiment of a tessellated spiral extraction mask overlaying an object of interest with inward radial progression commencing from a boundary point of the object of interest;
FIG. 10(a) illustrates a second embodiment of a tessellated spiral extraction mask overlaying an object of interest with outward radial progression commencing from the centroid of the object of interest;
FIG. 10(b) illustrates a second embodiment of a tessellated spiral extraction mask overlaying an object of interest with inward radial progression commencing from a boundary point of the object of interest;
FIG. 11(a) illustrates a third embodiment of a tessellated spiral extraction mask overlaying an object of interest with outward radial progression commencing from the centroid of the object of interest;
FIG. 11(b) illustrates a third embodiment of a tessellated spiral extraction mask overlaying an object of interest with inward radial progression commencing from a boundary point of the object of interest;
FIG. 12(a) illustrates a fourth embodiment of a tessellated spiral extraction mask overlaying an object of interest with outward radial progression commencing from the centroid of the object of interest;
FIG. 12(b) illustrates a fourth embodiment of a tessellated spiral extraction mask overlaying an object of interest with inward radial progression commencing from a boundary point of the object of interest;
FIG. 13(a) illustrates a fifth embodiment of a tessellated spiral extraction mask overlaying an object of interest with outward radial progression commencing from the centroid of the object of interest;
FIG. 13(b) illustrates a fifth embodiment of a tessellated spiral extraction mask overlaying an object of interest with inward radial progression commencing from a boundary point of the object of interest;
FIG. 14(a) illustrates a sixth embodiment of a tessellated spiral extraction mask overlaying an object of interest with outward radial progression commencing from the centroid of the object of interest;
FIG. 14(b) illustrates a sixth embodiment of a tessellated spiral extraction mask overlaying an object of interest with inward radial progression commencing from a boundary point of the object of interest;
FIG. 15(a) illustrates a seventh embodiment of a tessellated spiral extraction mask overlaying an object of interest with outward radial progression commencing from the centroid of the object of interest;
FIG. 15(b) illustrates a seventh embodiment of a tessellated spiral extraction mask overlaying an object of interest with inward radial progression commencing from a boundary point of the object of interest;
FIG. 16(a) illustrates an eighth embodiment of a tessellated spiral extraction mask overlaying an object of interest with outward radial progression commencing from the centroid of the object of interest;
FIG. 16(b) illustrates an eighth embodiment of a tessellated spiral extraction mask overlaying an object of interest with inward radial progression commencing from a boundary point of the object of interest;
FIG. 17(a) illustrates a ninth embodiment of a tessellated spiral extraction mask overlaying an object of interest with outward radial progression commencing from the centroid of the object of interest;
FIG. 17(b) illustrates a ninth embodiment of a tessellated spiral extraction mask overlaying an object of interest with inward radial progression commencing from a boundary point of the object of interest;
FIG. 18(a) illustrates a tenth embodiment of a tessellated spiral extraction mask overlaying an object of interest with outward radial progression commencing from the centroid of the object of interest;
FIG. 18(b) illustrates a tenth embodiment of a tessellated spiral extraction mask overlaying an object of interest with inward radial progression commencing from a boundary point of the object of interest;
FIG. 19(a) illustrates an eleventh embodiment of a tessellated spiral extraction mask overlaying an object of interest with outward radial progression commencing from the centroid of the object of interest;
FIG. 19(b) illustrates a eleventh embodiment of a tessellated spiral extraction mask overlaying an object of interest with inward radial progression commencing from a boundary point of the object of interest;
FIG. 20(a) illustrates a twelfth embodiment of a tessellated spiral extraction mask overlaying an object of interest with outward radial progression commencing from the centroid of the object of interest;
FIG. 20(b) illustrates a twelfth embodiment of a tessellated spiral extraction mask overlaying an object of interest with inward radial progression commencing from a boundary point of the object of interest;
FIG. 21(a) illustrates a thirteenth embodiment of a tessellated spiral extraction mask overlaying an object of interest with outward radial progression commencing from the centroid of the object of interest;
FIG. 21(b) illustrates a thirteenth embodiment of a tessellated spiral extraction mask overlaying an object of interest with inward radial progression commencing from a boundary point of the object of interest;
FIG. 22(a) illustrates a fourteenth embodiment of a tessellated spiral extraction mask overlaying an object of interest with outward radial progression commencing from the centroid of the object of interest;
FIG. 22(b) illustrates a fourteenth embodiment of a tessellated spiral extraction mask overlaying an object of interest with inward radial progression commencing from a boundary point of the object of interest;
FIG. 23(a) illustrates a fifteenth embodiment of a tessellated spiral extraction mask overlaying an object of interest with outward radial progression commencing from the centroid of the object of interest;
FIG. 23(b) illustrates a fifteenth embodiment of a tessellated spiral extraction mask overlaying an object of interest with inward radial progression commencing from a boundary point of the object of interest;
FIG. 24(a) illustrates a sixteenth embodiment of a tessellated spiral extraction mask overlaying an object of interest with outward radial progression commencing from the centroid of the object of interest;
FIG. 24(b) illustrates a sixteenth embodiment of a tessellated spiral extraction mask overlaying an object of interest with inward radial progression commencing from a boundary point of the object of interest;
FIG. 25(a) illustrates a seventeenth embodiment of a tessellated spiral extraction mask overlaying an object of interest with outward radial progression commencing from the centroid of the object of interest;
FIG. 25(b) illustrates a seventeenth embodiment of a tessellated spiral extraction mask overlaying an object of interest with inward radial progression commencing from a boundary point of the object of interest;
FIG. 26(a) illustrates an eighteenth embodiment of a tessellated spiral extraction mask overlaying an object of interest with outward radial progression commencing from the centroid of the object of interest;
FIG. 26(b) illustrates an eighteenth embodiment of a tessellated spiral extraction mask overlaying an object of interest with inward radial progression commencing from a boundary point of the object of interest;
FIG. 27(a) illustrates a nineteenth embodiment of a tessellated spiral extraction mask overlaying an object of interest with outward radial progression commencing from the centroid of the object of interest;
FIG. 27(b) illustrates a nineteenth embodiment of a tessellated spiral extraction mask overlaying an object of interest with inward radial progression commencing from a boundary point of the object of interest;
FIG. 28(a) illustrates an exemplary Voderberg spiral extraction mask;
FIG. 28(b) illustrates a twentieth embodiment of a tessellated spiral extraction mask overlaying an object of interest with outward radial progression commencing from the centroid of the object of interest;
FIG. 28(c) illustrates a twentieth embodiment of a tessellated spiral extraction mask overlaying an object of interest with inward radial progression commencing from a boundary point of the object of interest;
FIG. 29(a) illustrates a twenty-first embodiment of a tessellated spiral extraction mask overlaying an object of interest with outward radial progression commencing from the centroid of the object of interest;
FIG. 29(b) illustrates a twenty-first embodiment of a tessellated spiral extraction mask overlaying an object of interest with inward radial progression commencing from a boundary point of the object of interest;
FIG. 30(a) illustrates a twenty-second embodiment of a tessellated spiral extraction mask overlaying an object of interest with outward radial progression commencing from the centroid of the object of interest;
FIG. 30(b) illustrates a twenty-second embodiment of a tessellated spiral extraction mask overlaying an object of interest with inward radial progression commencing from a boundary point of the object of interest;
FIG. 31(a) illustrates a twenty-third embodiment of a tessellated spiral extraction mask overlaying an object of interest with outward radial progression commencing from the centroid of the object of interest;
FIG. 31(b) illustrates a twenty-third embodiment of a tessellated spiral extraction mask overlaying an object of interest with inward radial progression commencing from a boundary point of the object of interest;
FIG. 32(a) illustrates a twenty-fourth embodiment of a tessellated spiral extraction mask overlaying an object of interest with outward radial progression commencing from the centroid of the object of interest;
FIG. 32(b) illustrates a twenty-fourth embodiment of a tessellated spiral extraction mask overlaying an object of interest with inward radial progression commencing from a boundary point of the object of interest;
FIG. 33(a) illustrates a twenty-fifth embodiment of a tessellated spiral extraction mask overlaying an object of interest with outward radial progression commencing from the centroid of the object of interest;
FIG. 33(b) illustrates a twenty-fifth embodiment of a tessellated spiral extraction mask overlaying an object of interest with inward radial progression commencing from a boundary point of the object of interest;
FIG. 34(a) illustrates a twenty-sixth embodiment of a tessellated spiral extraction mask overlaying an object of interest with outward radial progression commencing from the centroid of the object of interest;
FIG. 34(b) illustrates a twenty-sixth embodiment of a tessellated spiral extraction mask overlaying an object of interest with inward radial progression commencing from a boundary point of the object of interest;
FIG. 35(a) illustrates a twenty-seventh embodiment of a tessellated spiral extraction mask overlaying an object of interest with outward radial progression commencing from the centroid of the object of interest;
FIG. 35(b) illustrates a twenty-seventh embodiment of a tessellated spiral extraction mask overlaying an object of interest with inward radial progression commencing from a boundary point of the object of interest;
FIG. 36(a) illustrates a twenty-eighth embodiment of a tessellated spiral extraction mask overlaying an object of interest with outward radial progression commencing from the centroid of the object of interest;
FIG. 36(b) illustrates a twenty-eighth embodiment of a tessellated spiral extraction mask overlaying an object of interest with inward radial progression commencing from a boundary point of the object of interest;
With reference to the figures, exemplary embodiments of the invention will now be described. The current invention is embodied in a technique for characterizing output data from an arrayed platform. While the invention is primarily described in connection with genomic microarrays, biochips and protein arrays, it will be appreciated by one of the ordinary skill in the art that the invention may be used in connection with a wide variety of arrayed platforms of detectors, including detectors for characterizing biological materials and other applications. It will also be appreciated that while the current invention is primarily described as a method, it may also be embodied in system comprising a computer processor and a memory coupled to the processor, where the memory is encoded with one or more programs that may perform the methods disclosed herein. One of ordinary skill in the art will also appreciate that the invention may also be embodied in an enhanced output pattern (which is an article of manufacture) that is the result of the techniques described herein.
With reference to
In an exemplary embodiment, the arrayed platform comprises an array of detectors or probes labeled with spectral emitting dyes (each referred to herein as an object of interest) useful in characterizing labeled genetic material within a sample (a microarray). The probes, sample or both may be pre-labeled. Preferably, the genetic material being analyzed is labeled using traditional techniques and dyes that may be detecting using fluorescence, chemiluminiscence, bioluminescence, and photoluminescence imaging modalities. Sample arrays include hybridized spotted cDNA microarrays, synthesized oligonucleotide arrays, spotted oligonucleotide arrays, peptide nucleotide assays, single nucleotide polymorphism (SNP) arrays, carbohydrate arrays, glycoprotein arrays, protein arrays, proteomic arrays, tissue arrays, antibody arrays, antigen arrays, bioassays, sequencing microarrays, sequencing by hybridization (SBH) microarrays, siRNA duplexes, RNAi arrays, glass-based arrays, nylon membrane arrays, thin film arrays, polymer-substrate arrays, capillary electrophoresis arrays, genospectral arrays, electronic arrays, bead arrays, quantum dot arrays, glycan arrays as well as any other arrays that are indicative of binding phenomenology and provide an output in a spectral regime.
The current invention relates to the analysis and characterization of a microarray output pattern. An output pattern is first examined to identify those pixels within the output pattern that correspond to objects of interest (this process is also known as registration). For example, see
In other embodiments, registration may occur using only information contained within the output pattern. For example, the output pattern may include explicit grid alignment controls/indicators present on the arrayed platform to facilitate the definition of corners, boundaries and guide points within the output pattern. By geo-referencing the grid alignment controls of the image, all objects of interest may be segmented. In other variations, registration takes place without any apriori information regarding the array by associating pixels within the output pattern to expected spot sizes, locations, geometric constraints, and other features of the array.
In output patterns where multiple dyes or multiple imaging modalities are employed, registration acts to correlate the differing modalities. For example cDNA arrays use Cy3 and Cy5 (two color dyes) as indicators of hybridization in differentially labeled samples, and as a result, two different output patterns are generated for two spectral dimensions. With such multiple dye arrangements, registration also correlates two different spots within the two images that represent the same physical detector probe location. Preferably, spatial overlaying, graph isomorphisms, sub-graph isomorphisms and polarization filters are used to overlay and register different images, and correct for local distortion effects.
In some embodiments, the current invention relates to the analysis of microarrays to detect the presence of genes and their respective expression levels. Such microarrays may include objects of interest such as genomic detectors including oligonucleotide, cDNA, RNA, and the output pattern represents the imaged intensity at the objects of interest on the microarray. Within the output pattern, the current invention also identifies pixels that are representative of each object of interest including their morphological representations and morphological invariants as related to the cumulative effects from source such as the substrate, probe type, microarray printing instrumentation, bioassay, and scanning instrumentation.
As described above, information from the instrument that created the array may be used as a guide such as spotter/genomic printer/arrayer layout files, spotter pin-head information, lithography masks, and mask generation files. In embodiments where output pattern is a representation of a capillary electrophoresis-based imaging array, the positioning, dimensions, and inter spacing of the capillary tubes, provide guiding information for downstream detection of post-hybridization intensity originating from a specific genomic element. Accordingly, prior to analysis, the pixels generally corresponding to objects of interest within the output pattern are identified through expected pre-determined properties.
If it is determined that the output pattern does not have sufficient contrast to identify pixels associated with an object of interest, various types of filters may be applied to the image. Such contrast enhancement may be useful in defining boundaries for the various objects of interest, to distinguish them from the background, that may then be used for generating segmentation contours. One embodiment convolves the output pattern with a Gabor filter, and more specifically with a set of two-dimensional Gabor filters.
The output pattern I_{spot }(x,y) , x, y∈Ω where Ω denotes a set of image points is convolved with a two-dimensional Gabor function
g(x,y), x,y∈Ω.
to obtain a Gabor feature image r(x,y) as follows:
One family of Gabor functions that may be used with the current invention is described below. The standard deviation σ of the Gaussian factor determines the effective size of the surrounding of a pixel in which weighted summation takes place. The eccentricity of the Gaussian and herewith the eccentricity of the convolution kernel g is determined by the parameter γ, called the spatial aspect ratio. Preferably, a value of γ=0.5 is used, although one of ordinary skill in the art will recognize that the spatial aspect ratio may be modified pursuant to the output pattern being characterized.
Pixels within the resulting convolved output pattern associated with an object of interest may then be extracted with a standard kernel convolution process. The extraction of such pixels progresses until a pre-defined condition is met indicating that the extraction is complete for a particular object of interest (i.e. a boundary condition). A standard edge detect kernel such as sobel operator may be used to generate edges that demarcate the pixels associated with an object of interest from pixels associated with the background. With some output patterns, it may be desirable to enhance the contrast therein. Alternative filtering techniques to Gabor filters that may be used with the current invention include: (i) low-pass/high-pass, band-pass filters; (ii) predetermined operators such as edge detection operators; (iii) Laplacian filters; and (iv) gradient-focusing filters.
Once the objects of interest are located, the current invention then optimally associates pixels with each object of interest to define an “extraction core” or “core”. For gene hybridization analysis the extraction core denotes the pixels that will be associated with an object of interest (i.e., a specific gene for genomic microarrays). All downstream decisions for detecting the absence or presence of an object of interest (e.g., gene), will be made based on this information contained within the core pixels as a starting basis. The parameters for defining a core may be:
(i) supplied by the user as expectation parameters. For example, in some embodiments the user provides his expectation of spot geometry, inter-spot spacing, spot diameter explicitly in terms of images pixels; or
(ii) provided by the instrumentation used to generate the array (i.e., spotter files, lithographic mask information). The resolution of the output pattern may also be utilized in combination with the spatial data to determine the optimal number of pixels to define the core. The scanning resolution (expressed in pixels terms—1 micron×1 micron pixel) is mapped to the feature size setting used by the spotter (expressed as 50 micron features or 18 micron features). This information is used to determine the number of pixels in the output image that capture an object of interest. Depending on the scanning resolution this may range from a single pixel to 10s or 1000s of pixels;
(iii) estimated using information-theoretic computations, if information regarding the core locations are unavailable. For example, an entropy map is generated based on intensity gradients computed over the entire image. A standard (3×3, 5×5, or N×N) sliding circular, polygonal or rectangular detection kernel is then applied to the entire image within the output pattern. The dimension N is so chosen to cover the largest expected spot or feature. The kernel application partitions or segments the image. An information entropy map is then computed based on the distance between extracted core geometry and the segmented feature. The arrayed image is then further segmented into regions with similar entropy distribution.
In some embodiments, a spatial filter is re-applied to the segmented objects (based on similar entropy measure), based on the array generation instrumentation configuration (i.e., spotter/pin head geometry) to then extract pixels associated with objects of interest. This arrangement requires an entropy map to be pre-computed for the objects of interest and then parameterized to accommodate variation due to artifacts and noise. This extraction process requires copying the core pixels (i.e., the pixels associated with a likely object of interest) to a data structure that can index to all the objects (denoting hybridized genes) associated with the array.
Once the core for each object of interest is identified, the segmented pixels associated with each such core are extracted. Once extracted, the segmented core pixels are used to populate a machine readable, data structure indexing the entire collection of objects of interest. Thereafter, the populated data structure is used to generate the modified output pattern comprising of the extracted pixels. The core extraction process is equivalent to deconstruction of the entire output array with m×n (denoted by I_{1}) pixels into a data structure that captures only the spatial intensities of the relevant objects and has pixels totaling
where j denotes the total number of objects of interest (e.g., pre-synthesized probes) on the array, and l_{ik }denotes the number of extracted pixels associate with the k-th feature of each object of interest. Also, k denotes the number of features encoding for object of interest i.
Also, I_{2}>I_{1}.
The current invention utilizes customized sophisticated traversal to compensate for sources of errors described above, all of which tend to reduce the signal ultimately present in the output pattern and/or increase the background noise or blur the boundary between the object of interest and background. As used herein, the term traversal refers to the process of how the segmented pixels associated with an object of interest are scanned by the algorithm for purposes of extracting and analyzing the core pixels (i.e., a hybridized spot). It is assumed that in images with low signal-to-noise (SNR) and signal-to-clutter ratio (SCR), or poor contrast, or saturation effects, segmentation will be imprecise and will necessarily include background pixels or exclude core pixels. As it is often difficult to predetermine the SNR or SCR without analyzing the entire image, or to identify those spots that have been degraded, the current invention assumes all images and all spots to be susceptible to suboptimal segmentation (as show in
In contrast to conventional microarray output analysis techniques, the pixels (which may be segmented) associated with the object of interest within the output pattern are spirally extracted to create a modified output pattern that primarily consists of those extracted pixels relating to objects of interest. For example, see
In some embodiments, as illustrated in FIGS. 6(a), the spiral extraction preferably uses spiral traversal 620 originating at center of the object of interest 620 (when known) which traverses outwardly 640 (as overlaid on grid 610), and in other embodiments an inward radial spiral traversal is utilized. As a further example, as illustrated in FIG. 6(b), a spiral transversal 670 overlaid on a grid 660 traverses outwardly from a centroid 675 to generate a sequence of extracted pixels 680. Spiral extractions are further illustrated in
With reference to FIGS. 8(a)-8(c), examples of outward logarithmic spiral traversals for extraction of images with varying parametric constants are illustrated. Each logarithmic spiral example 800, 840, 880 includes a spiral 820, 860, 890 overlaid on a grid 810, 840, 885 that traverses outwardly 830, 870, 895. The equation for the logarithmic spiral curve in polar coordinates is given by
R=ae^{bθ}
where r is the distance from the centroid, θ is the angle from the x-axis, and a and b are arbitrary parametric constants spiral curvature and modulation (as show in FIGS. 8(a)-8(c)). The logarithmic spiral family is also known as the growth spiral, equiangular spiral, and spiral mirabilis, and can be expressed parametrically as
x=r cosθ=a cosθe^{b0}
y=r sinθ=a sinθe^{b0}.
With reference to FIGS. 6(a) and 7 a graphical representation of an exemplary logarithmic spiral is illustrated. The logarithmic spiral may be constructed from equally spaced rays (straight lines intersecting two predetermined Cartesian coordinates) by starting at a point along one ray, and drawing the next ray perpendicular to a neighboring ray. As the number of rays approach infinity, the sequence of segments approaches a logarithmic spiral. Therefore a logarithmic spiral (also referred to as an equiangular spiral) is defined as a spiral that forms a constant angle between a line from the origin to any point on the curve and the angle of the tangent line at that point and its tangent is equal to the original angle. The family of equiangular spirals (including the Bernoulli spiral and logistique) are defined as spirals with a monotonic curve that cuts all radii vectors at a constant angle. The inverse of these spirals is also the same spiral. So the inverse spiral of an equiangular spiral defines an extraction trajectory to be implemented for adaptively extracting the segmented object of interest.
One equation for the rate of change of radius, for a spiral described by
is given by
and the angle between the tangent and radial line at the point ^{(r,θ) }is computed using:
So, as b→0, ^{ψ→π/}2 the spiral approaches a circle. A circle is also represented as a spiral for which the radius grows exponentially with the angle in the case of logarithmic spirals. The distances where a radius from the origin meets the curve are in geometric progression.
The arc length (s), curvature (κ), and tangential angle (φ) of the logarithmic spiral are computed using
With reference to FIG. 6(b), the morphing of a spiral trajectory to a circular trajectory as defined above is illustrated. At any extraction point the spiral and circular extractions are interchangeable. This is particularly important for implementing the extraction trajectory using a computer. Also, this behavior in the limit condition provides a mechanism for accurately extracting perfect or near-perfect circular cores using spiral extraction algorithms.
The parametric equations of a curve (f(t), g(t)) with radial point (x_{0}, y_{0}) and parameterized by a variable t are given by
Derivatives are then taken with respect to the parameter t. The software code stub in C++ language for implementing the procedure for extracting a logarithmic spiral is provided in the Software Appendix.
The current invention accommodates the extraction of objects of interest that are distributed geometrically irregular on the array. The output image deconstruction step, implemented via core extraction obviates the requirement for image geometry or distribution patterns. For example, when the boundaries of the objects of interest to be extracted have been pre-segmented, and the segmented objects have been registered and enumerated explicitly, the extraction algorithm traverses the object of interest in a spiral trajectory (outward logarithmic spiral or inward). This results in a linear vector of extracted pixels. As illustrated in
With reference to FIGS. 9(a)-9(c), an embodiment of a spiral extraction is illustrated with radial tilings. FIG. 9(a) illustrates an object of interest 910 overlaid on a pixellated grid 910 within a section of fluorescently labeled hybridized image obtained from a microarray output pattern generated using a confocal laser scanner operating at a specific wavelength and photomultiplier tube setting. The object of interest in the foreground of the output pattern can be extracted, as shown in FIG. 9(b) using an outward radial spiral extraction mask 920 with sectors of tessellated triangles converging at the centroid of the object of interest that traverse outwardly 930. FIG. 9(c) illustrates the extraction of the same segmented core extraction using the same radial spiral extraction mask 940 with inward progression that commences at any point on the segmented boundary or at the spiral mask template that traverses inwardly 950. All the sector extractions can be used for further processing or a plurality of sectors may be used.
With reference to FIGS. 10(a)-10(b), a variation of spiral extraction is illustrated where the tessellated triangles in a spiral mask template are not isosceles, and a radial layout is formed with a pair of adjacent sectors that are interlaid as mirror-images of each other. An extraction mask 1000, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral that traverses outwardly 1010 originating at the centroid 1015 of the object of interest (see FIG. 10(a)). In addition, an extraction mask 1020, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral progression that originates at any point on the boundary of the object of interest (see FIG. 10(b)) and traverses inwardly 1030 towards the centroid 1025 of the object of interest.
With reference to FIGS. 11(a)-11(b), another variation of spiral extraction is illustrated where the triangular wedges have varying angles, which do not always meet edge-to-edge, to generate a asymmetrical tessellation. An extraction mask 1100, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral that traverses outwardly 1110 originating at the centroid 1115 of the object of interest (see FIG. 11(a)). In addition, an extraction mask 1120, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral progression that originates at any point on the boundary of the object of interest (see FIG. 11(b)) and traverses inwardly 1130 towards the centroid 1125 of the object of interest.
With reference to FIGS. 12(a)-12(b), another embodiment of spiral extraction is illustrated where the tiling of 1×2 right triangles is not edge-to-edge (as compared to FIG. 11(a)) or made of equal sectors (as shown in FIG. 10(a)). An extraction mask 1200, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral that traverses outwardly 1210 originating at the centroid 1215 of the object of interest (see FIG. 12(a)). In addition, an extraction mask 1220, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral progression that originates at any point on the boundary of the object of interest (see FIG. 12(b)) and traverses inwardly 1230 towards the centroid 1225 of the object of interest.
With reference to FIGS. 13(a)-13(b), in some embodiments, the spiral extraction uses joining or modifying triangles which form rhombuses with radial tessellations. An extraction mask 1300, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral that traverses outwardly 1310 originating at the centroid 1315 of the object of interest (see FIG. 13(a)). In addition, an extraction mask 1320, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral progression that originates at any point on the boundary of the object of interest (see FIG. 13(b)) and traverses inwardly 1330 towards the centroid 1325 of the object of interest.
With reference to FIGS. 14(a)-14(b), a spiral extraction is illustrated that incorporates spiral tiling. This embodiment incorporates radial triangular tessellations with an even number of sectors that are bisected by straight lines. An extraction mask 1400, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral that traverses outwardly 1410 originating at the centroid 1415 of the object of interest (see FIG. 14(a)). In addition, an extraction mask 1420, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral progression that originates at any point on the boundary of the object of interest (see FIG. 14(b)) and traverses inwardly 1430 towards the centroid 1425 of the object of interest.
With reference to FIGS. 15(a)-15(b), another variation of a spiral extraction that utilizes spiral tiling is pictured. In this arrangement, two halves of a radial tiling on straight lines generate the spiral tiling. This form results in an offset among two sectors of identically-oriented triangles and creates a continuous “lane” of triangles. An extraction mask 1500, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral that traverses outwardly 1510 originating at the centroid 1515 of the object of interest (see FIG. 15(a)). In addition, an extraction mask 1520, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral progression that originates at any point on the boundary of the object of interest (see FIG. 15(b)) and traverses inwardly 1530 towards the centroid 1525 of the object of interest.
With reference to FIGS. 16(a)-16(b), an embodiment is disclosed where the tiling is offset by two cells where the lane of sectored triangles is wider than the previous embodiments. This type of tiling is useful for spots that exhibit local photobleaching artifacts and irregular probe distribution as they cannot be eliminated from the output pattern. If the artifacts are removed two other sectors of identically-oriented triangles are juxtaposed to create a new lane with different orientation. An extraction mask 1600, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral that traverses outwardly 1610 originating at the centroid 1615 of the object of interest (see FIG. 16(a)). In addition, an extraction mask 1620, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral progression that originates at any point on the boundary of the object of interest (see FIG. 16(b)) and traverses inwardly 1630 towards the centroid 1625 of the object of interest.
The extraction tilings discussed below with regard to
With reference to FIGS. 17(a)-17(b), an embodiment is shown with spiral tilings of non-isosceles triangle tiles. An extraction mask 1700, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral that traverses outwardly 1710 originating at the centroid 1715 of the object of interest (see FIG. 17(a)). In addition, an extraction mask 1720, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral progression that originates at any point on the boundary of the object of interest (see FIG. 17(b)) and traverses inwardly 1730 towards the centroid 1725 of the object of interest.
With reference to FIGS. 18(a)-18(b), an embodiment is disclosed where two triangular halves are offset to obtain a spiral tiling with kinked spirals. An extraction mask 1800, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral that traverses outwardly 1810 originating at the centroid 1815 of the object of interest (see FIG. 18(a)). In addition, an extraction mask 1820, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral progression that originates at any point on the boundary of the object of interest (see FIG. 18(b)) and traverses inwardly 1830 towards the centroid 1825 of the object of interest.
With reference to FIGS. 19(a)-19(b), an embodiment is disclosed where the tiling is offset by two cells. An extraction mask 1900, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral that traverses outwardly 1910 originating at the centroid 1915 of the object of interest (see FIG. 19(a)). In addition, an extraction mask 1920, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral progression that originates at any point on the boundary of the object of interest (see FIG. 19(b)) and traverses inwardly 1230 towards the centroid 1925 of the object of interest.
With reference to FIGS. 20(a)-20(b), an embodiment is disclosed where spiral tilings are formed from other radial tessellations with an even number of sectors. This embodiment commences with a rhomboidal tiling. An extraction mask 2000, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral that traverses outwardly 2010 originating at the centroid 2015 of the object of interest (see FIG. 20(a)). In addition, an extraction mask 2020, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral progression that originates at any point on the boundary of the object of interest (see FIG. 20(b)) and traverses inwardly 2030 towards the centroid 2025 of the object of interest.
With reference to FIGS. 21(a)-21(b), an embodiment is disclosed where two halves of a rhombus tiling are offset to obtain a spiral tiling. With this arrangement, the offset tilings juxtapose two sectors of identically-oriented rhombuses to creates a continuous “lane”. An extraction mask 2100, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral that traverses outwardly 2110 originating at the centroid 2115 of the object of interest (see FIG. 21(a)). In addition, an extraction mask 2120, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral progression that originates at any point on the boundary of the object of interest (see FIG. 21(b)) and traverses inwardly 2130 towards the centroid 2125 of the object of interest.
With reference to FIGS. 22(a)-22(b), an embodiment is disclosed with a tessellation comprised of deformed 45 degree triangles. As all the angles are 45 or 90 degrees, this tiling is especially well suited for segmented core with significant discontinuities. In this arrangement, the two alternating pairs of tiles are mirror images of interlaced pairs. An extraction mask 2200, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral that traverses outwardly 2210 originating at the centroid 2215 of the object of interest (see FIG. 22(a)). In addition, an extraction mask 2220, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral progression that originates at any point on the boundary of the object of interest (see FIG. 22(b)) and traverses inwardly 2230 towards the centroid 2225 of the object of interest.
With reference to FIGS. 23(a)-23(b), an embodiment is shown which is a variation of the embodiment of FIGS. 22(a)-22(b) except that the tiling has been offset by one tile. An extraction mask 2300, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral that traverses outwardly 2010 originating at the centroid 2315 of the object of interest (see FIG. 23(a)). In addition, an extraction mask 2320, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral progression that originates at any point on the boundary of the object of interest (see FIG. 23(b)) and traverses inwardly 2330 towards the centroid 2325 of the object of interest.
With reference to FIGS. 24(a)-24(b), an embodiment is disclosed where the tiling has been offset by two tiles to generate a “lane” of identically oriented tiles appears. An extraction mask 2400, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral that traverses outwardly 2410 originating at the centroid 2415 of the object of interest (see FIG. 24(a)). In addition, an extraction mask 2420, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral progression that originates at any point on the boundary of the object of interest (see FIG. 24(b)) and traverses inwardly 2430 towards the centroid 2425 of the object of interest.
With reference to FIGS. 25(a)-25(b), an embodiment is disclosed with tiling analogous to the one shown in FIGS. 24(a)-24(b) except the tiles are based on 30-degree triangles. An extraction mask 2500, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral that traverses outwardly 2510 originating at the centroid 2515 of the object of interest (see FIG. 25(a)). In addition, an extraction mask 2520, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral progression that originates at any point on the boundary of the object of interest (see FIG. 25(b)) and traverses inwardly 2530 towards the centroid 2525 of the object of interest.
With reference to FIGS. 26(a)-26(b), an embodiment is disclosed where the tiling of FIGS. 25(a)-25(b) has been offset by one tile. An extraction mask 2600, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral that traverses outwardly 2610 originating at the centroid 2615 of the object of interest (see FIG. 26(a)). In addition, an extraction mask 2620, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral progression that originates at any point on the boundary of the object of interest (see FIG. 26(b)) and traverses inwardly 2630 towards the centroid 2625 of the object of interest.
With reference to FIGS. 27(a)-27(b), an embodiment is disclosed where the tiling of FIGS. 26(a)-26(b) has been offset by two tiles. An extraction mask 2700, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral that traverses outwardly 2710 originating at the centroid 2715 of the object of interest (see FIG. 27(a)). In addition, an extraction mask 2720, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral progression that originates at any point on the boundary of the object of interest (see FIG. 27(b)) and traverses inwardly 2030 towards the centroid 2025 of the object of interest.
With reference to FIGS. 28(a)-28(c), embodiments are illustrated that use a Voderberg spiral extraction tiling technique for segmented core extraction. The unit tile, as shown in FIG. 28(a) is a bent enneagon. All lettered edges are equal in length. As two tiles fit together with two-fold symmetry, AB is parallel to EF, and DE is parallel to BC. Like most radial tiles, this arrangement can also be derived by deforming a triangle, in this case an isosceles 12-degree triangle. With reference to FIGS. 28(b)-28(c), an embodiment is disclosed where a Voderberg tile is used to construct a simple radial tiling. An extraction mask 2840, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral that traverses outwardly 2850 originating at the centroid 2845 of the object of interest (see FIG. 28(b)). In addition, an extraction mask 2860, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral progression that originates at any point on the boundary of the object of interest (see FIG. 28(c)) and traverses inwardly 2870 towards the centroid 2865 of the object of interest.
With reference to FIGS. 29(a)-29(b), an embodiment is shown which uses another variation of spiral Voderberg tilings. The pixellated hooks in the tiles enhance the spiral appearance and the small apical angle of the tiles makes for very smooth curves. This construct can be used to extract and preserve complex polygonal artifacts within segmented core or extracted and computational resampled cores. An extraction mask 2900, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral that traverses outwardly 2910 originating at the centroid 2915 of the object of interest (see FIG. 29(a)). In addition, an extraction mask 2920, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral progression that originates at any point on the boundary of the object of interest (see FIG. 29(b)) and traverses inwardly 2930 towards the centroid 2925 of the object of interest.
With reference to FIGS. 30(a)-30(b), a spiral extraction utilizing a bent wedge tiling is illustrated. This spiral extraction technique comprises an isosceles 15-degree triangle distorted into a curved enneagon. The apical angle is 15 degrees, the three successive angles on each side are 165 degrees, and the two end angles are 60 and 105 degrees. The 165 degree angles are also the interior angles of a 24-gon, meaning the intraspot tiles can fit together in a myriad of ways. As seen here, the tiles can form a simple radial pattern. As the tiles can curve in either direction within each annulus, this arrangement permits an infinite number of tilings. An extraction mask 3000, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral that traverses outwardly 3010 originating at the centroid 3015 of the object of interest (see FIG. 30(a)). In addition, an extraction mask 3020, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral progression that originates at any point on the boundary of the object of interest (see FIG. 30(b)) and traverses inwardly 3030 towards the centroid 3025 of the object of interest.
With reference to FIGS. 31(a)-31(b), an embodiment is disclosed where the bent-wedge tiling described above is offset in a variety of ways to create extraction spirals. In addition to the familiar ways of creating spirals, here, a 90-degree sector has been offset to make a spiral. An extraction mask 3100, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral that traverses outwardly 3110 originating at the centroid 3115 of the object of interest (see FIG. 31(a)). In addition, an extraction mask 3120, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral progression that originates at any point on the boundary of the object of interest (see FIG. 30(b)) and traverses inwardly 3130 towards the centroid 3125 of the object of interest.
With reference to FIGS. 32(a)-32(b), an extraction mask 3200, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral that traverses outwardly 3210 originating at the centroid 3215 of the object of interest (see FIG. 32(a)). In addition, an extraction mask 3220, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral progression that originates at any point on the boundary of the object of interest (see FIG. 32(b)) and traverses inwardly 3230 towards the centroid 3225 of the object of interest.
With reference to FIGS. 33(a)-33(b), an embodiment of an offset variation of the Hirschorn tiling of FIGS. 32(a)-32(b) is disclosed. An extraction mask 3300, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral that traverses outwardly 3310 originating at the centroid 3315 of the object of interest (see FIG. 33(a)). In addition, an extraction mask 3320, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral progression that originates at any point on the boundary of the object of interest (see FIG. 33(b)) and traverses inwardly 3330 towards the centroid 3325 of the object of interest.
With reference to FIGS. 34(a)-34(b), an embodiment is disclosed where the handedness of the central rosette and the next annulus of the tiling have been reversed and the radial angle is biased in the clockwise direction compared to the embodiment of FIGS. 33(a)-33(b). An extraction mask 3400, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral that traverses outwardly 3410 originating at the centroid 3415 of the object of interest (see FIG. 34(a)). In addition, an extraction mask 3420, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral progression that originates at any point on the boundary of the object of interest (see FIG. 34(b)) and traverses inwardly 3430 towards the centroid 3425 of the object of interest.
With reference to FIGS. 35(a)-35(b), a spiral extraction utilizing Hirschorn tiling is disclosed. This tiling consists of equilateral pentagons. In order, the vertex angles are 60, 160, 80, 100 and 140 degrees. The pentagon can be considered an equilateral triangle attached to a rhombus with 80-degree acute angles. Beyond the initial rosette, the tiling consists of radial sectors made up of pairs of mirror-image pentagons. The pairs stack in the usual triangular stacking pattern.. An extraction mask 3500, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral that traverses outwardly 3010 originating at the centroid 3515 of the object of interest (see FIG. 35(a)). In addition, an extraction mask 3520, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral progression that originates at any point on the boundary of the object of interest (see FIG. 35(b)) and traverses inwardly 3530 towards the centroid 3525 of the object of interest.
With reference to FIGS. 36(a)-36(b), an embodiment is disclosed where the handedness of the central rosette and the next annulus of the tiling have been reversed as compared to FIGS. 35(a)-35(b). An extraction mask 3600, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral that traverses outwardly 3010 originating at the centroid 3615 of the object of interest (see FIG. 36(a)). In addition, an extraction mask 3620, overlaid on object of interest 910 (and interposed on grid 900), is derived from a radial spiral progression that originates at any point on the boundary of the object of interest (see FIG. 36(b)) and traverses inwardly 3630 towards the centroid 3625 of the object of interest.
In some embodiments, the objects of interest are distributed geometrically irregular on the array. With these types of arrangements, the boundary of each object of interest is identified and the extraction algorithm traverses the object of interest in a spiral trajectory (outward logarithmic spiral or inward). This results in a linear vector of extracted pixels. As illustrated in
The following steps describe how a spiral extraction is configured to extract spatial intensity data from objects of interests having varying geometric sizes and distributed geometrically irregularly on the arrayed output image.
The arc length s, or a spiral, can be described in polar coordinates as
Depending upon the core size, one can calculate the spiral extraction length size by solving the differential equation, which results out of the relation y′=tan(b+F):
Substituting y=x·z and rewriting in polar coordinates gives the spiral's extraction equation.
Outward spiral traversals are accomplished by setting the segmented core centroid in center of the synthesis region to serve as a “seed” pixel. The spiral traverses from the centroid to segmented boundaries in an outward spiral using the equations described above for a logarithmic spiral.
Inward spiral transversals are accomplished by starting at the boundaries of the segmented core, traverse from the boundaries to the centroid in an inward spiral using the inverse spiral equations described above. Any point on the boundary of the segmented core can be used as “seed” pixel for the traversal. During extraction traversal, the spatial vector of pixels, in the order, in which they are traversed are copied to a linear array computer data structure. This is exemplified in the code provided in the Appendix.
The process of spiral extraction for outward radial spiral traversals includes the steps of marking the object centroid as the seed. Radially traversing outwards until all of the segment boundary pixels have been traversed and their outward neighboring pixels have also been traversed. Each traversed pixel is marked as traversed and its address is stored in a linear array along with the corresponding pixel intensity. Centroid coordinates mark the first address in the linear array data structure. This step ceases when the outward traversal is completed.
The process of spiral extraction for inward radial spiral traversal commences by marking the object centroid as a termination point. Extraction starts at any segment boundary pixel as the seed. The spiral traverses along the circumference of the object of interest in the anticlockwise or clockwise direction, until the entire object of interest is engulfed. The inward traversal can now be initiated using any segment pixel as the extraction seed. Once the extraction seed pixel is chosen, the linear array provides the pre-stored coordinates and spatial intensity corresponding to that pixel. Once the seed pixel has been found, the entire array of pixels down to the centroid, in the reverse order are stored within a newly created linear vector that will contain the extracted object of interest.
Similarly, for an outward radial spiral traversal, the centroid for the object of interest is marked as the seed pixel. The spiral radially traverses outwards until all the segment boundary pixels have been traversed and the outwardly adjacent pixels have also been traversed. Each traversed pixel is marked as traversed and its address is stored in a linear array along with the corresponding pixel intensity. The centroid coordinates mark the first address in the linear array data structure. The process ceases when the outward traversal is completed.
In most embodiments, a gradient measurement is utilized to determine how many pixels are required to adequately characterize an object of interest within the output pattern. A spatial gradient is computed in x and y directions from the core centroid. The second derivative of the spatial gradient is then computed to find hard discontinuities and boundary of the core to define the pixels associated with each core.
After the extraction of the pixels associated with an object of interest is completed, the extracted pixels are then transformed into the spectral domain to increase resolution. In some embodiments, the step of transforming the extracted pixels into the spectral domain commences with the step of computationally re-sampling the extracted pixels to increase the spatial resolution of the pixellated object. A two-dimensional surface interpolating using a set of scattered spatial data points is generated using either the method of Renka and Cline or a modification of Shepard's method.
Computational resealing provides increased spatial resolution and fine-grained morphologically for images that have limited dynamic range and/or high background noise. Computational resealing amplifies morphological artifacts thereby increasing opportunity for feature detection. This provides an opportunity for ultra-sensitive hybridization detection in genomic microarrays, when using very small amounts of starting material.
In the exemplary embodiment, an interpolating surface F(x, y) is constructed thromgh a set of m scattered data points (x_{r}, y_{r}, f_{r}), for r=1, 2, . . . , m. Also, in the (x, y) plane, the extracted pixels, corresponding to an object of interest, must be distinct. The constructed surface is continuous and has continuous first derivatives. Synthetic resampling involves firstly creating a triangulation with all the (x, y) data points as nodes, the triangulation being as nearly equiangular as possible. Then gradients in the x- and y-directions are estimated at node r, for r=1, 2, . . . , m, as the partial derivatives of a quadratic function of x and y which interpolates the data value f_{r}, and which fits the data values at nearby nodes (those within a fixed distance provided as an input to the algorithm) in a weighted least-squares sense. The weights are chosen such that closer nodes have more influence than more distant nodes on derivative estimates at node r.
The computed partial derivatives, with the f_{r }values, at the three nodes of each triangle define a piecewise polynomial surface of a certain form which is the interpolant on that triangle. The interpolant F(x, y) can subsequently be evaluated at n points (x_{k}, y_{k}), for k=1, 2, . . . , n, inside or outside the domain of the data by points outside the domain are evaluated by extrapolation.
An alternate technique that can be used for synthetic resampling is based on modification of Shepard's method. The basic Shepard method, interpolates the input data with the weighted mean. The basic method is global in that the interpolated value at any point depends on all the data. The method is made local by adjusting each w_{r}(x, y) to be zero outside a circle with centre (x_{r}, y_{r}) and some radius R_{w}. Also, to improve the performance of the basic method, each f_{r }above is replaced by a function f_{r}(x, y), which is a quadratic fitted by weighted least-squares to data local to (x_{r}, y_{r}) and forced to interpolate (x_{r}, y_{r}, f_{r}). In this context, a point (x, y) is defined to be local to another point if it lies within some distance R_{q }of it. Computation of this quadratic constitutes the bulk of computational effort. If there are less than 5 other points within distance R_{q }from (x_{r}, y_{r}), the quadratic is replaced by a linear function. In cases of rank-deficiency, the minimum norm solution is computed.
The interpolant F(x, y) can subsequently be evaluated at any point (x, y) inside or outside the domain of the data. Points outside the domain are evaluated by extrapolation. Traditionally this method has been used in adapting grid size in adaptive finite element (FEM) analysis and computer graphics. Use to microarray resampling is new.
As demonstrated, computational re-sampling facilitates the preservation of feature consistency over a four log dilution range down to 1 ng concentration level.
The next step transforms the extracted core pixels into the spectral domain, using the Fast Fourier Transform (FFT). In general the FFT provides the means of transforming a signal defined in the time domain into one defined in the frequency domain. Core extraction leads to a spatial signal vector of length N. The transformed spectra are also of length N. A length N sequence x can be denoted by x(n), n=0,1,2, . . . N−1 where x(n) may be real (x∈R^{N}) or complex (x∈C^{N}).
This is achieved as follows. Consider the core x as a vector x in an N dimensional vector space. That is, each extraction core x(n) is regarded as a coordinate in that space. Also this spatial vector, denoted as x is mathematically a single point in N-space represented by a list of coordinates (x_{0}, x_{z}, x_{2}, . . . , x_{N−1}). It can be interpreted geometrically as an arrow in N-space from the origin 0 Δ(0,0, . . . ,0) to the point x Δ(x_{0}, x_{1}, x_{2}, . . . , x_{N−1}).
We define the following as equivalent:
xΔ x Δx(·)Δ(x_{0}, x_{1}, . . . , x_{N−1})Δ[x_{0}, x_{1}, . . . , x_{N−1}]Δ[x_{0}, x_{1}, . . . , x_{N−1}]
where x_{n} Δx(n) is the nth sample of the spatial core extraction x.
For purposes of exemplary embodiment consider all extraction vectors to be of length N.
The current invention utilizes zero padding for extraction cores with less than N components (say for small spots). Zero padding consists of appending zeros to a signal. It maps a length N signal to a length M>N signal, but M need not be an integer multiple of N:
For example,
Z
The above definition is natural when x(n) represents a signal starting at time 0 and extending for N samples. In general, consider a complex series x(k) with N samples of the form
x_{0}, x_{2}, x_{3 }. . . x_{k }. . . x_{N−1}
where x is a complex number
X _{i} =X _{real} +J X _{imaq}
Further, assume that that the series outside the range 0, N−1 is extended N-periodic, that is, X_{k}=x_{k+N }for all k. The FFT of this series will be denoted X(k), it will also have N samples. The forward transform will be defined as
The Discrete Fourier Transform (DFT) of a signal x may be defined by:
where
Δ means “is defined as” or “equals by definition”
x_{p}(t_{n}) Δ input signal amplitude (real or complex) at time t_{n }(sec)
t_{n } Δ nT=nth sampling instant (sec)
n Δ sample number (integer)
T Δ sampling period (sec)
X_{p}(ω_{k}) Δ Spectrum of x_{p }(complex amplitude), at radian frequency ω_{k }(rad/sec)
ω_{k } Δ kΩ=kth frequency sample (rad,/sec)
f_{S } Δ 1/T=sampling rate (samples/sec, or Hertz (Hz))
N=number of samples in both time and frequency (integer)
DFT can be rewritten
where x(n) denotes the input signal at time (sample) n, and X(k) denotes the kth spectral sample.
This general definition above for a complex series can be also defined for a real valued series. A real valued series can be represented by setting the imaginary part to 0. In general, the transform into the frequency domain will be a complex valued function, that is, with magnitude and phase.
The above equations are used for transforming a spatial core to a spectral core.
The spectral core is further converted to a power series representation using the Welch's method. Welch's method (or the periodogram method) for estimating power spectral density is carried out by dividing the time signal into successive blocks, and averaging squared-magnitude DFTs of the signal blocks. Let denote the x_{m }denote the mth block of the signal x, and let M denote the number of blocks. Then the PSD estimate is given by
The power spectral density representation R_{x}(k) is used for characterizing and analyzing the extraction core.
In some embodiments, the current invention also utilizes the step of convolving the extracted pixels with discretized coefficients of a function (linear functions, non-linear functions, canonical kernel functions.) In general, the following relationships hold:
Also, the convolution operation between two spectra x and y in C^{N }is denoted “x*y” and defined by
The basic concept is that a window of some finite size and shape is scanned across the spectrally transformed vector. The window with its weights is called the convolution kernel. If the filter h[j,k] is zero outside the (rectangular) window {j=0,1, . . . ,J−1; k=0,1, . . . ,K−1}, then the convolution can be written as the following finite sum:
The general steps for convolving the spectral core with a convolution kernel are given by:
(a) Compute A(Ω,Ψ)=F{a[m,n]}
(b) Multiply A(Ω,Ψ) by the precomputed (Ω,Ψ)=F{h[m,n]}
(c) Compute the result c[m,n]=F^{−1}{A(Ω,Ψ)*(Ω,Ψ)}
where f, g, and h are arbitrary functions and a is a constant. Convolution satisfies the properties
f*g=q*f
f*(g*h)=(f*q)*h
f*(g+h)=(f*g)+(f*h)
The property, f*(g*h), of successive convolutions is also denoted as a convolution cascade. In general, it can be extended to k convolutions, f*(g*(h*(i* . . . (*k))) . . . ))) The above relationships are used to increase the spatial resolution by convolution of the spectrally transformed pixels within extraction core.
With reference to
In the exemplary embodiment, the following operations are utilized to enhance the spectral resolution of extraction core. First, the spatial core of length N is partitioned into M sub-cores of length J (where J<N). The portioned cores may or may not be overlapping in the spatial domain and the lengths of the sub-cores do not need to be identical. Second, each spatial core of length J is then transformed to a spectral core using FFT described above. Third, the spectral cores of length J are then convolved with 1 discrete kernels where 1>0 with each of the sub-cores being convolved with the same convolution kernel, or in the alternative, each sub-core is convolved with a different convolution kernel. The convolution cascades applied to sub-cores can be of the form of: (i) a serial cascade: with no-recombination of the convolution results; (ii) a parallel cascade which comprises the recombination of the convolution results in the spectral domain after 1 or more convolutions; or (iii) a combination of serial and parallel Cascades which comprises the application of series of convolutions, followed by re-combinations, followed by further convolution cascades.
The extracted pixels that have been transformed into the spectral domain are useful as a stand-alone dataset to characterize the analyzed array as extraneous information not related to an object of interest has been removed through the extraction process and the resolution of the extracted pixels have been increased through spectral domain transformation. In addition, the transformed extracted pixels may also be useful for other applications including parameterizing a pre-determined dynamical system, parameterizing a pre-determined distributed representation of a single canonical dynamical system.
In some embodiments, it may be desirable to partition the extracted core of pixels if the spectral components therein are from a polycyclostationary or a cyclostationary system which represented by k sinusoids. With regard to
The exemplary embodiments have been primarily described with reference to diagrams illustrating pertinent components of the embodiments. It should be appreciated that not all components of a complete implementation of a practical system are necessarily illustrated or described in detail, nor are all of the varying component layout schema described. Rather, only those components and architectures necessary for a thorough understanding of the invention have been illustrated and described in detail. Actual implementations may contain more components or, depending upon the implementation, fewer components. Modifications to the preferred embodiments will be apparent to those skilled in the art. For example, the extraction and re-scaling techniques described above which compensate for low SCR or low SNR are applicable to processing a broad class of spatial and spatio-temporal images. It will also be appreciated that the extraction techniques and spectral transformation techniques described herein are independently useful and need not be practiced together. Consequently, the scope of the present invention should not be limited by the particular embodiments discussed above, but should be defined only by the claims set forth below and equivalents thereof.
Cited Patent | Filing date | Publication date | Applicant | Title |
---|---|---|---|---|
US5371669 | Jun 18, 1992 | Dec 6, 1994 | The United States Of America As Represented By The Administrator Of The National Aeronautics And Space Administration | Sliding mode control method having terminal convergence in finite time |
US5371834 | Aug 28, 1992 | Dec 6, 1994 | The United States Of America As Represented By The Administrator Of The National Aeronautics And Space Administration | Adaptive neuron model--an architecture for the rapid learning of nonlinear topological transformations |
US5631734 | Feb 10, 1994 | May 20, 1997 | Affymetrix, Inc. | Method and apparatus for detection of fluorescently labeled materials |
US5733729 | Sep 14, 1995 | Mar 31, 1998 | Affymetrix, Inc. | Computer-aided probability base calling for arrays of nucleic acid probes on chips |
US5784162 | Dec 12, 1995 | Jul 21, 1998 | Applied Spectral Imaging Ltd. | Spectral bio-imaging methods for biological research, medical diagnostics and therapy |
US5812272 * | Jan 30, 1997 | Sep 22, 1998 | Hewlett-Packard Company | Apparatus and method with tiled light source array for integrated assay sensing |
US5822466 | May 27, 1994 | Oct 13, 1998 | Max-Planck-Gesselschaft Zur Forderburg Der Wissenschaften E. V | Method and means of spatial filtering |
US5858659 | Nov 29, 1995 | Jan 12, 1999 | Affymetrix, Inc. | Polymorphism detection |
US5925525 | Apr 3, 1998 | Jul 20, 1999 | Affymetrix, Inc. | Method of identifying nucleotide differences |
US5968740 | Mar 5, 1997 | Oct 19, 1999 | Affymetrix, Inc. | Method of Identifying a Base in a Nucleic Acid |
US5974164 | Oct 16, 1995 | Oct 26, 1999 | Affymetrix, Inc. | Computer-aided visualization and analysis system for sequence evaluation |
US6025601 | Jun 9, 1997 | Feb 15, 2000 | Affymetrix, Inc. | Method and apparatus for imaging a sample on a device |
US6066454 | Oct 10, 1997 | May 23, 2000 | Affymetrix, Inc. | Computer-aided probability base calling for arrays of nucleic acid probes on chips |
US6090555 | Dec 23, 1997 | Jul 18, 2000 | Affymetrix, Inc. | Scanned image alignment systems and methods |
US6136541 | Feb 22, 1999 | Oct 24, 2000 | Vialogy Corporation | Method and apparatus for analyzing hybridized biochip patterns using resonance interactions employing quantum expressor functions |
US6142681 | Feb 22, 1999 | Nov 7, 2000 | Vialogy Corporation | Method and apparatus for interpreting hybridized bioelectronic DNA microarray patterns using self-scaling convergent reverberant dynamics |
US6165734 * | Nov 20, 1998 | Dec 26, 2000 | Applied Spectral Imaging Ltd. | In-situ method of analyzing cells |
US6171793 | Apr 19, 1999 | Jan 9, 2001 | Affymetrix, Inc. | Method for scanning gene probe array to produce data having dynamic range that exceeds that of scanner |
US6185561 | Jul 15, 1999 | Feb 6, 2001 | Affymetrix, Inc. | Method and apparatus for providing and expression data mining database |
US6218122 | Jun 16, 1999 | Apr 17, 2001 | Rosetta Inpharmatics, Inc. | Methods of monitoring disease states and therapies using gene expression profiles |
US6223127 | Aug 14, 1998 | Apr 24, 2001 | Affymetrix, Inc. | Polymorphism detection utilizing clustering analysis |
US6225625 | Jun 1, 1995 | May 1, 2001 | Affymetrix, Inc. | Signal detection methods and apparatus |
US6228593 | Jan 14, 2000 | May 8, 2001 | Affymetrix, Inc. | Computer-aided probability base calling for arrays of nucleic acid probes on chips |
US6229911 | Jul 24, 1998 | May 8, 2001 | Affymetrix, Inc. | Method and apparatus for providing a bioinformatics database |
US6242180 | Sep 23, 1998 | Jun 5, 2001 | Affymetrix, Inc. | Computer-aided visualization and analysis system for sequence evaluation |
US6245511 | Feb 22, 1999 | Jun 12, 2001 | Vialogy Corp | Method and apparatus for exponentially convergent therapy effectiveness monitoring using DNA microarray based viral load measurements |
US6294327 | Jan 26, 1998 | Sep 25, 2001 | Affymetrix, Inc. | Apparatus and method for detecting samples labeled with material having strong light scattering properties, using reflection mode light and diffuse scattering |
US6308170 | Jul 24, 1998 | Oct 23, 2001 | Affymetrix Inc. | Gene expression and evaluation system |
US6342355 | Jan 5, 2000 | Jan 29, 2002 | The United States Of America As Represented By The Department Of Health & Human Services | Probe-based analysis of heterozygous mutations using two-color labelling |
US6344316 | Jun 25, 1997 | Feb 5, 2002 | Affymetrix, Inc. | Nucleic acid analysis techniques |
US6349144 * | Feb 7, 1998 | Feb 19, 2002 | Biodiscovery, Inc. | Automated DNA array segmentation and analysis |
US6361937 | Mar 19, 1996 | Mar 26, 2002 | Affymetrix, Incorporated | Computer-aided nucleic acid sequencing |
US6368799 | Jun 11, 1998 | Apr 9, 2002 | Affymetrix, Inc. | Method to detect gene polymorphisms and monitor allelic expression employing a probe array |
US6370478 | Dec 28, 1998 | Apr 9, 2002 | Rosetta Inpharmatics, Inc. | Methods for drug interaction prediction using biological response profiles |
US6391550 | May 27, 1999 | May 21, 2002 | Affymetrix, Inc. | Identification of molecular sequence signatures and methods involving the same |
US6420108 | Feb 9, 1998 | Jul 16, 2002 | Affymetrix, Inc. | Computer-aided display for comparative gene expression |
US6453241 | Dec 23, 1998 | Sep 17, 2002 | Rosetta Inpharmatics, Inc. | Method and system for analyzing biological response signal data |
US6468744 | Jan 2, 1998 | Oct 22, 2002 | Affymetrix, Inc. | Analysis of genetic polymorphisms and gene copy number |
US6485913 * | Oct 2, 2000 | Nov 26, 2002 | Sequenom, Inc. | Systems and methods for performing reactions in an unsealed environment |
US6490533 | Dec 3, 2001 | Dec 3, 2002 | Affymetrix, Inc. | System, method, and product for dynamic noise reduction in scanning of biological materials |
US6495363 * | May 18, 2001 | Dec 17, 2002 | Orchid Biosciences, Inc. | In-line complete spectral fluorescent imaging of nucleic acid molecules |
US6498863 * | Sep 20, 2000 | Dec 24, 2002 | Media Cybernetics Inc. | Method, system, and product for analyzing a digitized image of an array to create an image of a grid overlay |
US6542638 | Feb 21, 2001 | Apr 1, 2003 | Shannon Roy Campbell | Method for matching spatial patterns |
US6567750 | Oct 20, 2000 | May 20, 2003 | Imaging Research, Inc. | Process for evaluating chemical and biological assays |
US6571005 | Apr 21, 2000 | May 27, 2003 | The Regents Of The University Of California | Feature extraction and normalization algorithms for high-density oligonucleotide gene expression array data |
US6671625 | Feb 17, 2000 | Dec 30, 2003 | Vialogy Corp. | Method and system for signal detection in arrayed instrumentation based on quantum resonance interferometry |
US6704662 | Jul 2, 2002 | Mar 9, 2004 | Vialogy Corp. | Technique for quantiating biological markers using quantum resonance interferometry |
US6716629 * | Oct 10, 2001 | Apr 6, 2004 | Biotrove, Inc. | Apparatus for assay, synthesis and storage, and methods of manufacture, use, and manipulation thereof |
US20010033364 * | Jan 3, 2001 | Oct 25, 2001 | Dario Cabib | Spectral bio-imaging of the eye |
US20020052882 | Jul 9, 2001 | May 2, 2002 | Seth Taylor | Method and apparatus for visualizing complex data sets |
US20030044775 | Jul 2, 2002 | Mar 6, 2003 | Vialogy Corp. | Technique for quantiating biological markers using quantum resonance interferometry |
US20030071843 | Oct 17, 2001 | Apr 17, 2003 | Bruce Hoff | System and method for specifying and applying microarray data preparation |
US20030112446 * | Oct 28, 2002 | Jun 19, 2003 | Benjamin Miller | Method for biomolecular sensing and system thereof |
US20030219150 * | May 24, 2002 | Nov 27, 2003 | Niles Scientific, Inc. | Method, system, and computer code for finding spots defined in biological microarrays |
US20030235840 * | Jan 24, 2003 | Dec 25, 2003 | Ward David C. | Multiparametric fluorescence in situ hybridization |
EP1162572A2 * | Jun 6, 2001 | Dec 12, 2001 | Agilent Technologies, Inc. (a Delaware corporation) | Method and system for evaluating orientation of molecular array |
Reference | ||
---|---|---|
1 | * | Cheung et al., Analysis of gene microarray images, 1999, pp. 627-632. |
2 | * | Kuklin, et al., High throughput screening of gene expression signature, 2000, pp. 41-46. |
3 | * | PCT/US03/14182 to VIOLOGY Corp. |
Citing Patent | Filing date | Publication date | Applicant | Title |
---|---|---|---|---|
US7352373 * | Sep 30, 2003 | Apr 1, 2008 | Sharp Laboratories Of America, Inc. | Systems and methods for multi-dimensional dither structure creation and application |
US7567876 | Aug 7, 2007 | Jul 28, 2009 | Vialogy Llc | Quantum resonance interferometry for detecting signals |
US7907271 | Jun 13, 2008 | Mar 15, 2011 | Historx, Inc. | Method and system for standardizing microscope instruments |
US7978258 | Jul 12, 2011 | Historx, Inc. | Automatic exposure time selection for imaging tissue | |
US8009889 * | Jun 27, 2007 | Aug 30, 2011 | Affymetrix, Inc. | Feature intensity reconstruction of biological probe array |
US8027030 | Sep 27, 2011 | Historx, Inc. | Method and system for standardizing microscope instruments | |
US8120768 | Jan 20, 2011 | Feb 21, 2012 | Historx, Inc. | Method and system for standardizing microscope instruments |
US8121365 | Aug 7, 2008 | Feb 21, 2012 | Historx, Inc. | Method and system for determining an optimal dilution of a reagent |
US8160348 | Aug 5, 2008 | Apr 17, 2012 | Historx, Inc. | Methods and system for validating sample images for quantitative immunoassays |
US8204705 | Jun 19, 2012 | Vialogy Llc | Quantum resonance interferometry for detecting signals | |
US8335360 | Dec 18, 2012 | Historx, Inc. | Compartment segregation by pixel characterization using image data clustering | |
US8369596 | Feb 5, 2013 | Affymetrix, Inc. | Feature intensity reconstruction of biological probe array | |
US8655037 | Dec 14, 2012 | Feb 18, 2014 | Historx, Inc. | Compartment segregation by pixel characterization using image data clustering |
US8934689 | Oct 30, 2012 | Jan 13, 2015 | Affymetrix, Inc. | Feature intensity reconstruction of biological probe array |
US8969255 | Nov 21, 2012 | Mar 3, 2015 | The Arizona Board Of Regents | Compound arrays for sample profiling |
US9147103 * | Nov 21, 2014 | Sep 29, 2015 | Affymetrix, Inc. | Feature intensity reconstruction of biological probe array |
US9240043 | Sep 15, 2009 | Jan 19, 2016 | Novartis Ag | Reproducible quantification of biomarker expression |
US20050068463 * | Sep 30, 2003 | Mar 31, 2005 | Sharp Laboratories Of America, Inc. | Systems and methods for multi-dimensional dither structure creation and application |
US20070269818 * | Dec 26, 2006 | Nov 22, 2007 | Affymetrix, Inc. | Carbohydrate arrays |
US20080033672 * | Aug 7, 2007 | Feb 7, 2008 | Vialogy Corp. | Quantum Resonance Interferometry for Detecting Signals |
US20080232657 * | Jun 27, 2007 | Sep 25, 2008 | Affymetrix, Inc. | Feature Intensity Reconstruction of Biological Probe Array |
US20080309929 * | Jun 13, 2008 | Dec 18, 2008 | Historx, Inc. | Method and system for standardizing microscope instruments |
US20090034823 * | May 14, 2008 | Feb 5, 2009 | Historx, Inc. | Compartment segregation by pixel characterization using image data clustering |
US20090074266 * | Aug 7, 2008 | Mar 19, 2009 | Historx, Inc. | Method and system for determining an optimal dilution of a reagent |
US20090086046 * | Aug 29, 2008 | Apr 2, 2009 | Historx, Inc. | Automatic exposure time selection for imaging tissue |
US20100023290 * | Jan 28, 2010 | Vialogy Llc | Quantum Resonance Interferometry for Detecting Signals | |
US20100136549 * | Sep 15, 2009 | Jun 3, 2010 | Historx, Inc. | Reproducible quantification of biomarker expression |
US20110116086 * | May 19, 2011 | Historx, Inc. | Method and system for standardizing microscope instruments | |
US20110116087 * | May 19, 2011 | Historx, Inc. | Method and system for standardizing microscope instruments | |
US20150098637 * | Nov 21, 2014 | Apr 9, 2015 | Affymetrix, Inc. | Feature Intensity Reconstruction of Biological Probe Array |
CN102749290A * | Jul 2, 2012 | Oct 24, 2012 | 浙江大学 | Method for detecting growth state of branches of crown canopy of cherry tree |
U.S. Classification | 382/133 |
International Classification | G06F19/20, G06K9/00, G06T5/00, G06T7/00, G01N21/25 |
Cooperative Classification | G06T7/0081, G06T2207/10056, G01N21/253, G06T2207/10064, G06T2207/30072, G06T7/0083, G06T2207/20156, G06T7/0012, G06T2207/20168, G06K9/00, G06F19/20 |
European Classification | G06T7/00S1, G06T7/00B2, G01N21/25B2, G06K9/00, G06T7/00S2 |
Date | Code | Event | Description |
---|---|---|---|
May 5, 2003 | AS | Assignment | Owner name: VIALOGY CORP., CALIFORNIA Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:GULATI, SANDEEP;REEL/FRAME:014054/0751 Effective date: 20030505 |
Oct 1, 2004 | AS | Assignment | Owner name: BIOPROJECTS INTERNATIONAL PLC, UNITED KINGDOM Free format text: SECURITY AGREEMENT;ASSIGNOR:VIALOGY CORP.;REEL/FRAME:015201/0844 Effective date: 20040709 |
Aug 28, 2009 | FPAY | Fee payment | Year of fee payment: 4 |
Oct 11, 2012 | AS | Assignment | Owner name: VIALOGY LLC, CALIFORNIA Free format text: MERGER;ASSIGNORS:ORIGINAL INVESTMENTS PLC;ORIGINAL INVESTMENTS LLC;VIALOGY CORP.;REEL/FRAME:029115/0893 Effective date: 20060926 |
Aug 28, 2013 | FPAY | Fee payment | Year of fee payment: 8 |